Re: [Json] Call for Consensus: Proposed Text for "8.1 Character Encoding"

[Nico Williams <[e-mail]>]

On Mon, Mar 13, 2017 at 10:12:09PM +0100, Carsten Bormann wrote:
> > 1) Does the working group think adding any text on how to detect the
> > encoding worthwhile?
> 
> No, that would be a regression into maintaining the fiction that
> UTF-16 and UTF-32 versions of JSON are being used in interchange.

I would support saying that JSON texts SHOULD be UTF-8.  I might even
support that they MUST be UTF-8, but I suspect there won't be consensus
for that, though IF there is, then I agree with you.

> Not sure if I’m allowed to note that after saying no above, but not

Why wouldn't you be?

> all JSON documents have four bytes.

True!  The empty string at the top-level, with no whitespace around it,
is a two-byte text when encoded in UTF-8.

Indeed, just two bytes will do for some cases; at most four are needed.
Byte order can be detected as well, also in two to four bytes, but byte
order also is pointless to think about if UTF-8 were REQUIRED (which it
isn't [yet]).

Nico
--